Textbook of Usability

Alert fatigue occurs when the volume of alerts exceeds the user's capacity to evaluate them, leading to habitual dismissal of all alerts — including the critical ones. It is the single most significant usability problem in clinical decision support systems, but also affects dashboards, security tools, and any interface that uses alerts liberally.

Studies consistently find that clinicians override 49–96% of clinical decision support (CDS) alerts. A typical hospital physician may receive hundreds of alerts per day, the vast majority of which are clinically insignificant. When most alerts are false alarms or irrelevant, the rational response is to dismiss all of them quickly — including the rare ones that matter.

The solution is not to add more alerts but to improve alert specificity:

• Reduce total alerts by eliminating those with low clinical significance
• Reserve interruptive modalities (pop-ups, hard stops) for genuinely critical situations
• Use tiered alerting: passive information → non-interruptive indicators → interruptive alerts → hard stops
• Measure and reduce override rates through alert refinement
• Personalise thresholds to reduce false positives

A well-designed CDS system generates fewer alerts, not more. Every alert added to the system should be evaluated not only for whether it catches a real problem but for its impact on the signal-to-noise ratio of the overall alert stream. Alert fatigue is the failure mode of alert-based monitoring, and it is common enough that alert design must anticipate and prevent it.